Message ID | 20190430071446.13716-1-xzhou@redhat.com |
---|---|
State | Changes Requested |
Delegated to: | Petr Vorel |
Headers | show |
Series | syscalls/swapon02: Do not fail on overlayfs | expand |
On Tue, Apr 30, 2019 at 3:21 PM Murphy Zhou <xzhou@redhat.com> wrote: > Currently swapfiles on Overlayfs are not supported. > > So if we are on overlayfs and we get EINVAL from swapon() we return TCONF. > > Signed-off-by: Murphy Zhou <xzhou@redhat.com> > --- > testcases/kernel/syscalls/swapon/swapon02.c | 5 +++++ > 1 file changed, 5 insertions(+) > > diff --git a/testcases/kernel/syscalls/swapon/swapon02.c > b/testcases/kernel/syscalls/swapon/swapon02.c > index 4af5105c6..211cdfc4e 100644 > --- a/testcases/kernel/syscalls/swapon/swapon02.c > +++ b/testcases/kernel/syscalls/swapon/swapon02.c > @@ -86,6 +86,11 @@ static void verify_swapon(struct test_case_t *test) > return; > } > > + if (fs_type == TST_OVERLAYFS_MAGIC && errno == EINVAL) { > + tst_resm(TCONF, "Swapfile on overlayfs not implemeted"); > + return; > + } > The code looks correct. But it already has a test skipping for BTRFS, is there any possibility to combine these filesystems check together? > + > tst_resm(TFAIL, "swapon(2) failed to produce expected error:" > " %d, errno: %s and got %d.", test->exp_errno, > test->exp_errval, TEST_ERRNO); > -- > 2.21.0 > > > -- > Mailing list info: https://lists.linux.it/listinfo/ltp >
On Tue, Apr 30, 2019 at 04:14:23PM +0800, Li Wang wrote: > On Tue, Apr 30, 2019 at 3:21 PM Murphy Zhou <xzhou@redhat.com> wrote: > > > Currently swapfiles on Overlayfs are not supported. > > > > So if we are on overlayfs and we get EINVAL from swapon() we return TCONF. > > > > Signed-off-by: Murphy Zhou <xzhou@redhat.com> > > --- > > testcases/kernel/syscalls/swapon/swapon02.c | 5 +++++ > > 1 file changed, 5 insertions(+) > > > > diff --git a/testcases/kernel/syscalls/swapon/swapon02.c > > b/testcases/kernel/syscalls/swapon/swapon02.c > > index 4af5105c6..211cdfc4e 100644 > > --- a/testcases/kernel/syscalls/swapon/swapon02.c > > +++ b/testcases/kernel/syscalls/swapon/swapon02.c > > @@ -86,6 +86,11 @@ static void verify_swapon(struct test_case_t *test) > > return; > > } > > > > + if (fs_type == TST_OVERLAYFS_MAGIC && errno == EINVAL) { > > + tst_resm(TCONF, "Swapfile on overlayfs not implemeted"); > > + return; > > + } > > > > The code looks correct. > > But it already has a test skipping for BTRFS, is there any possibility to > combine these filesystems check together? Good idea~ Sending V2. Thansk! M > > > > + > > tst_resm(TFAIL, "swapon(2) failed to produce expected error:" > > " %d, errno: %s and got %d.", test->exp_errno, > > test->exp_errval, TEST_ERRNO); > > -- > > 2.21.0 > > > > > > -- > > Mailing list info: https://lists.linux.it/listinfo/ltp > > > > > -- > Regards, > Li Wang
On Tue, Apr 30, 2019 at 4:31 PM Murphy Zhou <xzhou@redhat.com> wrote: > On Tue, Apr 30, 2019 at 04:14:23PM +0800, Li Wang wrote: > > On Tue, Apr 30, 2019 at 3:21 PM Murphy Zhou <xzhou@redhat.com> wrote: > > > > > Currently swapfiles on Overlayfs are not supported. > > > > > > So if we are on overlayfs and we get EINVAL from swapon() we return > TCONF. > > > > > > Signed-off-by: Murphy Zhou <xzhou@redhat.com> > > > --- > > > testcases/kernel/syscalls/swapon/swapon02.c | 5 +++++ > > > 1 file changed, 5 insertions(+) > > > > > > diff --git a/testcases/kernel/syscalls/swapon/swapon02.c > > > b/testcases/kernel/syscalls/swapon/swapon02.c > > > index 4af5105c6..211cdfc4e 100644 > > > --- a/testcases/kernel/syscalls/swapon/swapon02.c > > > +++ b/testcases/kernel/syscalls/swapon/swapon02.c > > > @@ -86,6 +86,11 @@ static void verify_swapon(struct test_case_t *test) > > > return; > > > } > > > > > > + if (fs_type == TST_OVERLAYFS_MAGIC && errno == EINVAL) { > > > + tst_resm(TCONF, "Swapfile on overlayfs not > implemeted"); > > > + return; > > > + } > > > > > > > The code looks correct. > > > > But it already has a test skipping for BTRFS, is there any possibility to > > combine these filesystems check together? > > Good idea~ Sending V2. > One more comment. I just noticed that it also has an FS skipping list in setup(), do u think can we move BTRFS and OVERLAYFS to there?
On Tue, Apr 30, 2019 at 04:54:23PM +0800, Li Wang wrote: > On Tue, Apr 30, 2019 at 4:31 PM Murphy Zhou <xzhou@redhat.com> wrote: > > > On Tue, Apr 30, 2019 at 04:14:23PM +0800, Li Wang wrote: > > > On Tue, Apr 30, 2019 at 3:21 PM Murphy Zhou <xzhou@redhat.com> wrote: > > > > > > > Currently swapfiles on Overlayfs are not supported. > > > > > > > > So if we are on overlayfs and we get EINVAL from swapon() we return > > TCONF. > > > > > > > > Signed-off-by: Murphy Zhou <xzhou@redhat.com> > > > > --- > > > > testcases/kernel/syscalls/swapon/swapon02.c | 5 +++++ > > > > 1 file changed, 5 insertions(+) > > > > > > > > diff --git a/testcases/kernel/syscalls/swapon/swapon02.c > > > > b/testcases/kernel/syscalls/swapon/swapon02.c > > > > index 4af5105c6..211cdfc4e 100644 > > > > --- a/testcases/kernel/syscalls/swapon/swapon02.c > > > > +++ b/testcases/kernel/syscalls/swapon/swapon02.c > > > > @@ -86,6 +86,11 @@ static void verify_swapon(struct test_case_t *test) > > > > return; > > > > } > > > > > > > > + if (fs_type == TST_OVERLAYFS_MAGIC && errno == EINVAL) { > > > > + tst_resm(TCONF, "Swapfile on overlayfs not > > implemeted"); > > > > + return; > > > > + } > > > > > > > > > > The code looks correct. > > > > > > But it already has a test skipping for BTRFS, is there any possibility to > > > combine these filesystems check together? > > > > Good idea~ Sending V2. > > > > One more comment. > > I just noticed that it also has an FS skipping list in setup(), do u think > can we move BTRFS and OVERLAYFS to there? Great! That makes perfect sense. Also I'm thinking a few other tweaks needed: NFS now actually support swapfiles. Let's remove it from skip list; Skip all swapon/swapoff tests on TMPFS BTRFS OVERLAYFS; Add CIFS_MAGIC, then skip on it too; Remove existing BTRFS return value tweaks. Thanks, M > > -- > Regards, > Li Wang
On Tue, Apr 30, 2019 at 5:08 PM Murphy Zhou <xzhou@redhat.com> wrote: > ... > > > > > + if (fs_type == TST_OVERLAYFS_MAGIC && errno == EINVAL) { > > > > > + tst_resm(TCONF, "Swapfile on overlayfs not > > > implemeted"); > > > > > + return; > > > > > + } > > > > > > > > > > > > > The code looks correct. > > > > > > > > But it already has a test skipping for BTRFS, is there any > possibility to > > > > combine these filesystems check together? > > > > > > Good idea~ Sending V2. > > > > > > > One more comment. > > > > I just noticed that it also has an FS skipping list in setup(), do u > think > > can we move BTRFS and OVERLAYFS to there? > > Great! That makes perfect sense. Also I'm thinking a few other tweaks > needed: > > NFS now actually support swapfiles. Let's remove it from skip list; > But AFAIK someone still run LTP on old distro/kernel, so we can't guarantee the test works fine in that situation. So, my opinion is to reserve the NFS, or do you have a better way to make test more flexible? > Skip all swapon/swapoff tests on TMPFS BTRFS OVERLAYFS; > Add CIFS_MAGIC, then skip on it too; > Remove existing BTRFS return value tweaks. > Agree.
On Tue, Apr 30, 2019 at 05:21:34PM +0800, Li Wang wrote: > On Tue, Apr 30, 2019 at 5:08 PM Murphy Zhou <xzhou@redhat.com> wrote: > > > ... > > > > > > + if (fs_type == TST_OVERLAYFS_MAGIC && errno == EINVAL) { > > > > > > + tst_resm(TCONF, "Swapfile on overlayfs not > > > > implemeted"); > > > > > > + return; > > > > > > + } > > > > > > > > > > > > > > > > The code looks correct. > > > > > > > > > > But it already has a test skipping for BTRFS, is there any > > possibility to > > > > > combine these filesystems check together? > > > > > > > > Good idea~ Sending V2. > > > > > > > > > > One more comment. > > > > > > I just noticed that it also has an FS skipping list in setup(), do u > > think > > > can we move BTRFS and OVERLAYFS to there? > > > > Great! That makes perfect sense. Also I'm thinking a few other tweaks > > needed: > > > > NFS now actually support swapfiles. Let's remove it from skip list; > > > > But AFAIK someone still run LTP on old distro/kernel, so we can't guarantee > the test works fine in that situation. So, my opinion is to reserve the > NFS, or do you have a better way to make test more flexible? Fair enough. We can reserve NFS on the skip list to avoid much false alarms. > > > > Skip all swapon/swapoff tests on TMPFS BTRFS OVERLAYFS; > > Add CIFS_MAGIC, then skip on it too; > > Remove existing BTRFS return value tweaks. > > > > Agree. Thanks! M > > -- > Regards, > Li Wang
On Tue, Apr 30, 2019 at 5:21 AM Li Wang <liwang@redhat.com> wrote: > > > > On Tue, Apr 30, 2019 at 5:08 PM Murphy Zhou <xzhou@redhat.com> wrote: >> >> ... >> > > > > + if (fs_type == TST_OVERLAYFS_MAGIC && errno == EINVAL) { >> > > > > + tst_resm(TCONF, "Swapfile on overlayfs not >> > > implemeted"); >> > > > > + return; >> > > > > + } >> > > > > >> > > > >> > > > The code looks correct. >> > > > >> > > > But it already has a test skipping for BTRFS, is there any possibility to >> > > > combine these filesystems check together? >> > > >> > > Good idea~ Sending V2. >> > > >> > >> > One more comment. >> > >> > I just noticed that it also has an FS skipping list in setup(), do u think >> > can we move BTRFS and OVERLAYFS to there? >> >> Great! That makes perfect sense. Also I'm thinking a few other tweaks >> needed: >> >> NFS now actually support swapfiles. Let's remove it from skip list; > > > But AFAIK someone still run LTP on old distro/kernel, so we can't guarantee the test works fine in that situation. So, my opinion is to reserve the NFS, or do you have a better way to make test more flexible? Avoid whitelist. Test for FIBMAP ioctl support. Thanks, Amir.
On Tue, Apr 30, 2019 at 05:37:36AM -0400, Amir Goldstein wrote: > On Tue, Apr 30, 2019 at 5:21 AM Li Wang <liwang@redhat.com> wrote: > > > > > > > > On Tue, Apr 30, 2019 at 5:08 PM Murphy Zhou <xzhou@redhat.com> wrote: > >> > >> ... > >> > > > > + if (fs_type == TST_OVERLAYFS_MAGIC && errno == EINVAL) { > >> > > > > + tst_resm(TCONF, "Swapfile on overlayfs not > >> > > implemeted"); > >> > > > > + return; > >> > > > > + } > >> > > > > > >> > > > > >> > > > The code looks correct. > >> > > > > >> > > > But it already has a test skipping for BTRFS, is there any possibility to > >> > > > combine these filesystems check together? > >> > > > >> > > Good idea~ Sending V2. > >> > > > >> > > >> > One more comment. > >> > > >> > I just noticed that it also has an FS skipping list in setup(), do u think > >> > can we move BTRFS and OVERLAYFS to there? > >> > >> Great! That makes perfect sense. Also I'm thinking a few other tweaks > >> needed: > >> > >> NFS now actually support swapfiles. Let's remove it from skip list; > > > > > > But AFAIK someone still run LTP on old distro/kernel, so we can't guarantee the test works fine in that situation. So, my opinion is to reserve the NFS, or do you have a better way to make test more flexible? > > Avoid whitelist. > Test for FIBMAP ioctl support. Hmm.. This is better. I'll try. Thanks! M > > Thanks, > Amir.
diff --git a/testcases/kernel/syscalls/swapon/swapon02.c b/testcases/kernel/syscalls/swapon/swapon02.c index 4af5105c6..211cdfc4e 100644 --- a/testcases/kernel/syscalls/swapon/swapon02.c +++ b/testcases/kernel/syscalls/swapon/swapon02.c @@ -86,6 +86,11 @@ static void verify_swapon(struct test_case_t *test) return; } + if (fs_type == TST_OVERLAYFS_MAGIC && errno == EINVAL) { + tst_resm(TCONF, "Swapfile on overlayfs not implemeted"); + return; + } + tst_resm(TFAIL, "swapon(2) failed to produce expected error:" " %d, errno: %s and got %d.", test->exp_errno, test->exp_errval, TEST_ERRNO);
Currently swapfiles on Overlayfs are not supported. So if we are on overlayfs and we get EINVAL from swapon() we return TCONF. Signed-off-by: Murphy Zhou <xzhou@redhat.com> --- testcases/kernel/syscalls/swapon/swapon02.c | 5 +++++ 1 file changed, 5 insertions(+)